Our research focuses on the reliability assessment of engineering systems using advanced computational methods, including active learning for efficient failure probability estimation, dynamic reliability modeling based on condition monitoring data, quantum computing for solving intractable reliability problems, and likelihood-free Bayesian computation to predict system degradation and reliability from noisy and limited data.

Active Learning for Structural Reliability: We focus on enhancing the reliability assessment of engineering structures through advanced active learning techniques. The framework integrates various methods, such as, surrogate modeling, Monte Carlo simulation, and adaptive learning. Surrogate models such as kriging and polynomial chaos expansions are employed to efficiently approximate the limit-state surface, making the exploration of the random space more cost-effective. 

Reliability Analysis of Complex Systems: Our research focuses on the reliability, life prediction, and health management of complex engineering systems. We develop dynamic reliability analysis methods and data-driven models using Bayesian networks and machine learning for health monitoring, fault detection, and remaining useful life prediction. This work supports reliability-based design, predictive maintenance strategies, and overall performance improvement, reducing downtime and ensuring operational efficiency in a variety of challenging environments.

Quantum Computing for Reliability Simulation: We explore the application of quantum computing to reliability engineering by developing quantum algorithms to address complex reliability problems, including the use of quantum fault trees. Our goal is to leverage the unique capabilities of quantum computing to solve reliability challenges that are intractable with classical methods. 

Bayesian Degradation Modeling and Prediction: We develop advanced likelihood-free algorithms for Bayesian degradation modeling and prediction, focusing on model selection and parameter estimation from noisy and limited condition monitoring data. By incorporating randomness and uncertainty, we try to simulate degradation processes over time, enabling life prediction and reliability assessment.